Are short sellers incrementally informed prior to earnings announcements?

Contrary to the hypothesis that informed short sellers increase their positions prior to earnings announcements, we find that short activity declines in the pre-announcement period compared with activity in non-announcement time. This statistically significant, but economically modest, decline may suggest that the fraction of informed short sellers actually increases if (as Diamond and Verrecchia (1987) suggest) the uncertainty around earnings announcements increases short selling costs and causes uninformed short sellers to withdraw from the market. While we find a statistically and economically significant inverse relation between pre-announcement short activity and announcement period returns, when we control for the non-announcement ability of short sellers to predict future returns documented by Diether et al. (2009), the significance of the relation between pre-announcement short activity and announcement period returns vanishes. Thus, we infer that short sellers are not incrementally informed prior to earnings announcements. ► We examine whether informed short sellers trade prior to earnings releases. ► Short selling is abnormally low prior to announcements. ► Pre-announcement short selling does not contain an abnormal amount of information.